The Correlation Between PID Temperature Setting and Water

Brewing Temperature for Gaggia Coffee Deluxe

By Auber Instrument
August 2007

Introduction
The water brewing temperature (Tb) can become much more consistent in the
Gaggia Espresso machine with the PID temperature controller modification. The
boiler temperature can be stabilized by a PID temperature controller in less than
10 minutes; however, the water brewing temperature takes longer time to
stabilize. The PID temperature setting (set value or SV) for the boiler should be
set higher than the optimal water brewing temperature to compensate the heat
loss of the grouphead. This study is to find the correlation between the PID
temperature setting and water brew temperature. It will help the user to make for
consistent espresso.

Equipments and set up

The espresso machine used for this study is a Gaggia Coffee Deluxe
manufactured in July 2005. The serial number is 0105114347. This machine has
the same boiler design that used on most Gaggia home espresso machines.
The PID kit used is the Auber PID Temperature Control Retrofit KIT for Gaggia.
Model number KIT-GG.
Water brewing temperature measurement was done by the Scace Thermofilter.
This device is currently used by World Barista Championship (WBC) and United
States Barista Competition (USBC) technical standards teams for testing and
evaluating the machines used in competitions. The details of the Scace
Thermofilter can be found at
http://www.home-barista.com/forums/scace-thermofilter-temperature-device-
instructions-t515.html
To make the Thermofilter fit the Gaggia Coffee, the bottom of Gaggia portafilter
was cut off so that the Thermofilter can be mounted.
The Thermofilter was connected to a digital thermometer that converts the T type
thermocouple signal to a 0-5 VDC analog signal output (Auber instrument, SYL-
1712). The thermometer was connected to a 16 bit data acquisition system. The
sampling rate was 128 Hz. One limitation of this system is that the output of the
thermometer has only 1 degree resolution.
All tests were performed in an air-conditioned rooms maintained at around 25 C
(76-77 F).

Calibration of the meter

The temperature standard we used is a precision mercury filled glass

thermometer. Temperature range is -20 to 150 °C with 1 degree division. This
meter is not NIST traceable but was calibrated with the ice water mixture and
boiling water technique described at

1
http://www.biggreenegg.com/boilingPoint.htm#pressure. It was also verified by a
NIST traceable thermometer at 37.0 C. The maximum error of this thermometer
is <0.2 C over the 0 to 110 C range.

Controller calibration

The set up for calibrating the controllers is shown in picture 1. A stainless steel
pitcher filled with cooking oil was placed on a hot plate with magnetic stirrer. The
hot plate was controlled by a PID temperature control system to make the
temperature stable. The mercury thermometer was inserted to the oil bath at its
immersion line. The temperature slowly changed from 80 C to 110 C at rate of
<0.5C/min. The controller reading was compared with the mercury thermometer,
with the error adjusted by the meter’s offset. After calibration, the controller had <
0.2 C errors when compared with mercury thermometer.

Picture 1, Set up for calibrating controllers.

Thermofilter calibration

Since the Thermofilter was not used for over 100C and the oil bath might
damage the glue used on it, it was calibrated in a Dewar thermos as shown in
Picture 2. The thermos was also used because it took a long time to equilibrate
the temperature, due to the large thermo mass attached and limit heat circulation
(this is not the case when measuring the brew water because the water moves
very fast). Boiling water was poured into the Dewar that contained the
Thermofilter and a calibrated RTD sensor. After 20 minutes of equilibration (cap
sealed), the reading was compared with the calibrated thermometer. After
calibration, the error of the Thermofilter was <0.3C.

2
 Picture 2. Thermos used for calibrating Thermofilter

After the experiment, we re-calibrated all the controllers and the Thermofilter to
make sure no instruments drifted over this period of time.

Flow rate of the Thermofilter

While the pump for Machine was running, we collected water from Thermofilter
for 120 seconds. We collect the water in the middle of pumping to eliminate the
error due to the dead volume in the Thermofilter. We collected 305.4 ml water
(measured by weight with a 0.1g resolution lab grade scale). That averages to a
flow rate of 2.55ml/s.

Results and discussions

The relationship between Tb and SV
215

210
Water Brewing Temp (F)

205

Average
200
Peak

195

190

185
205 210 215 220 225 230
Set Temperature (F)

3
 102

100

Water Brewing Temp. (C)

98

96

94 Average
92 Peak

90

88

86

84
98 100 102 104 106 108 110
Set Temperature (C)

Figure 1. Correlation between the PID temperature controller set temperature

(SV) and the water brewing temperature (Tb) detected at the grouphead by
Scace Thermofilter. Top, Display in degrees Fahrenheit. Bottom, Display in
degrees Celsius.

Figure 1 shows the relationship between Tb and SV. All data was taken 30
minutes after the SV is changed and after several shots to let system stabilized.
The brew water temperature is not constant during the shot. Their waveform is
shown in Fig. 2. The peak value can be 10 C (18F) higher than the temperature
at the end of shot. That is due to the small size of the boiler and the material
used for the boiler (Aluminum). Because the waveform changes at different SV,
we think using the average temperature instead of peak will be more consistent
and closer to the real espresso making situation.

0.98

0.96

0.94 103
99
Temp x100 (C)

0.92 100
102
0.9
106
0.88 109
107
0.86
101
0.84

0.82

0.8
1
Time (4s/div)

4
Figure 2. The waveform for of the temperature change at different SV. Each trace
represents a 25 seconds double shot. The SV value (in Celsius) is listed on the
right side of the graph. Only one group of tests used in Fig. 1 is presented in this
figure.

Time required for brew temperature to stabilize.

SV=216 F

205

200
Temp (F)

195
Avg
Max
190

185

180
0 10 20 30 40 50 60
Time (min)

SV=102 C

96
94
92
Temp (C)

90 Avg
88 Max

86
84
82
0 10 20 30 40 50 60
Time (min)

Figure 3. Brew water temperature changes as the machine warms up. Each data
point represents a 25 second double shot. Both the peak temperature (Max) and
the average temperature during the brewing are plotted. Top, plot in degrees
Fahrenheit. Bottom, Plot in degrees Celsius.

5
Figures 3 shows the water brewing temperature (Tb) change measured by the
Scace Thermofilter versus time. The timers started when the machine was
powered up from room temperature. Each data point represents a 25 second
double shot shown Figure 4. Both the average temperature and the peak
temperature are presented. The data is collected at minimum of 4 minutes
interval.

From these figures we can see that when the machine started from cold, the Tb
can be 9 degree C lower than when it is warmed up, even if the boiler
temperature was stabilized. Therefore, it is important to let the machines warm
up before brewing starts. After the machine was powered up for 20 minutes, the
Tb stabilized to +/- 2 degree C.

SV=102 C

0.95

49
0.9
42
34
Temp (x100 C)

30
0.85
26
22
0.8 18
14
11
0.75 7

0.7
1
Time (4 s/div)

Figure 4. The waveforms used to generate the data points used in Fig. 3. The
legend on the right side is time in minutes. It represents the time since the
machine powered up.

Time required between each shot

6
 SV=216 F

210
Brew Water Temp (F)
205

200
Average
195
Peak
190

185

180
0 5 10 15 20
Time Interval (min)

SV=102 C

98
Brew Water Temp (C)

96
94
92
Average
90
Peak
88
86
84
82
0 2 4 6 8 10 12 14 16
Time Interval (min)

Figure 5. Brew water temperature of the second shot as a time function from the
first shot. Top, plot in degrees Fahrenheit. Bottom, Plot in degrees Celsius.

Figure 5 shows the temperature of the second shot at different time interval from
the first shot. In this study, we set the SV at 216 F (102 C) and let the machine
stabilized for 1 hour before start the testing. Then, we made two consequential
shots with different time interval between the shots. The first shot was stabilized
for at least 10 minutes before the second shot. We measured brew water
temperature of the second shot only. The time zero is when the first shot started.
This study shown that although the controller can stabilized the temperature near
its sensor location in less than 2 minutes (data not shown), the brew water
temperature will take at least 4 minutes to recover to ± 2 C of its steady state. 7
to 8 minutes is needed for temperature to be in the ± 1C from the steady state.

7
Conclusion
When making a double shot (60 ml), the brew water temperature of Gaggia
Coffee deluxe can drop 10 C (18 F) at the end of the shot. However the
temperature change profile is reproducible and predictable. When correctly
operated, the Gaggia can produce a very reproducible shot. The brew
temperature will take at least 20 minutes to stabilize after start from room
temperature. The time needed between the shots should be at least 4 minutes or
longer. The optimal PID setting temperature should be between 210 to 225 F (99
to 107 C)

Acknowledgement
The author would like to thank Praveen Nadaraju for many helpful suggestions
on this study.